EP0273504B1

EP0273504B1 - Circuit for extracting carrier signals

Info

Publication number: EP0273504B1
Application number: EP87202438A
Authority: EP
Inventors: Dirk Muilwijk
Original assignee: Philips Gloeilampenfabrieken NV; Koninklijke Philips Electronics NV
Current assignee: Koninklijke Philips NV
Priority date: 1986-12-08
Filing date: 1987-12-07
Publication date: 1993-03-10
Anticipated expiration: 2007-12-07
Also published as: NL8603110A; DK638687A; DK168843B1; US4890302A; CA1273680A; DK638687D0; EP0273504A1; DE3784675T2; JPS63160448A; DE3784675D1; JP2541588B2; AU8214187A

Description

The invention relates to a carrier reproducing circuit for extracting the carrier signal from a pulse-coded modulated carrier signal, this circuit comprising a clock pulse regenerator, a multiplier having a multiplication factor N(N = 2, 3, ...) for generating an N^th harmonic from the modulated carrier signal, a first bandpass filter coupled to the multiplier for isolating the aforesaid N^th harmonic, a sampling circuit which is inserted between the multiplier and the first bandpass filter and which is connected to the clock pulse regenerator for sampling the N^th harmonic under the control of a clock pulse signal originating from the clock pulse regenerator at no less than one predetermined sampling point per symbol time T of the data signal and a divider having the dividend N which is connected to the first bandpass filter.
Such a carrier reproducing circuit is known from US-A 3,835,404. This document describes a carrier reproducing circuit for PSK signals wherein the jitter due to phase displacements in the carrier signal occurring at bit transitions is avoided by sampling with clock pulses having a pulse-width which is smaller than the distance between the successive phase displacements in the modulated carrier signal but having as large a finite width as possible.
It is known to make use of continuous phase modulation of carrier signal because carrier signals modulated in such a way have a constant amplitude enabling non-linear amplification, and because they usually have a small bandwidth. This modulation category comprises among others Tamed Frequency Modulation (TFM), Quadrivalent Three-bit Correlated Cosinusoidal Modulation (Q3RC) and Correlative Phase Shift Keying (CORPSK).
It has appeared that a large jitter value commonly occurs in a carrier signal extracted from such a pulse-coded continuous phase-modulated carrier signal.
The invention has for its object to strongly reduce in a simple manner the jitter effect occurring when reproducing a carrier signal from a continuous phase-modulated carrier signal.
The circuit for extracting carrier signals in accordance with the invention is characterized in that the pulse coded modulated carrier signal is a pulse coded continuous phase modulated carrier signal having a rational modulation index and that the sampling circuit provides in an instantaneous sampling of the N^th harmonic under the control of the clock pulse signal.
The invention is based on the insight that in the case of continuous phase modulated carrier signal for extracting the carrier signal from such a continuous phase-modulated carrier signal it is essential that instantaneous sampling at predetermined sampling points be achieved for obtaining as reliable a reproduced carrier signal as possible.
It is observed that a carrier reproducing circuit for extracting the carrier signal from a pulse-coded continuous phase -modulated carrier signal is known inter alia from the article entitled "Synchronization properties of continuous phase modulation " by J.Aulin and C.E. Sundberg, published in the "Conference Papers" of "Globecom 82, Global Telecommunication Conference, Miami, 29th November - 2nd December 1982. However in this known circuit no sampling circuit is provided between the multiplier and the first bandpass filters.
The invention also relates to a receiver comprising a carrier reproducing circuit for extracting the carrier signal from a pulse-coded modulated carrier signal, this circuit comprising a clock pulse regenerator, a multiplier having a multiplication factor N(N = 2, 3, ...) for generating an N^th harmonic from the modulated carrier signal, a first bandpass filter coupled to the multiplier for isolating the aforesaid N^th harmonic, a sampling circuit which is inserted between the multiplier and the first bandpass filter and which is connected to the clock pulse regenerator for sampling the N^th harmonic under the control of a clock pulse signal originating from the clock pulse regenerator at no less than one predetermined sampling point per symbol time T of the data signal and a divider having the dividend N which is connected to the first bandpass filter characterized in that the pulse coded modulated carrier signal is a pulse coded continuous phase modulated carrier signal having a rational modulation index and that the sampling circuit (7) provides in a instantaneous sampling of the N^th harmonic under the control of the clock pulse signal.
The invention and its advantages will further be explained with reference tot he embodiments shown in the drawing, wherein corresponding elements in the Figures are indicated by identical reference numerals and wherein:

Figure 1:: shows a block diagram of a receiver comprising a carrier reproducing circuit for continuous phase-modulated signals having a rational modulation index in accordance with the invention,
Figure 2a:: shows the impulse response and
Figure 2b:: shows the waveform of a 2RC-signal,
Figure 3a:: shows the impulse response and
Figure 3b:: shows the waveform for CORPSK (2-3, 1 + D) and duebinary CPM with h = 0.5 and
Figure 4a:: shows the impulse response and
Figure 4b:: shows the waveform for TFM and TFSK, respectively.

With the receiver comprising a carrier-reproducing circuit 10, shown in Figure 1, it is possible to extract in a highly reliable way a continuous phase-modulated carrier signal (CPM-signal) applied to input-terminal 1 and having a rational modulation index h such that the resulting carrier has a small phase jitter. Before embarking on the explanation of the carrier reproducing circuit, the structure of such a CPM-signal is further elucidated.
An information signal to be transferred digitally, represented by a series (a_m) of m-nary digital symbols,

${(a}_{m} {) = ..a}_{-2} {,a}_{-1} {,a₀,a₁,a₂, ..a}_{n} (1)$

with $a_{m} =±(2n+1)$
and n=0,1,2,...etc. can be represented as a function of time by

${a(t) = Σa}_{m} p(t=mT) (2)$

with T representing the duration of the digital symbols and p(t) representing a rectangular pulse of a length of T.
Before transferring such signals they are preferably modulated in that the modulated carrier signal has a constant amplitude. The major advantage of such a carrier signal is the fact that non-linear signal processing such as non-linear amplification for obtaining a high efficiency, can be used without affecting the information contained in the signal
Each digital modulation having a constant amplitude can be written as:

${u(t) = cos(ω}_{c} t + φ(t)) (3)$

wherein ω_c represents the angular frequency of the carrier signal and φ(t) represents the phase as a function of time.
If φ(t) is varied as a function of a (t) the following equation will be obtained:

$φ(t) = φ{a(t)) (4)$
In Phase Shift Keying (PSK) φ (t) is kept constant over the symbol time T and changed abruptly at the symbol transitions.
However, there are also types of modulation such as Continuous Phase Frequency Shift Keying (CPFSK), for which the following condition holds:

$d φ(t)/d(t) = K.a(t) (5)$

wherein K is a constant. With continuous Phase Frequency Shift Keying the detection process for obtaining a better S/N ratio can be spread out over more than one symbol interval.
Examples of CPFSK are Minimum Shift Keying (MSK) and Sinusoidal Frequency Shift Keying (SFSK). MSK is a modulation entailing a linear phase shift of ± π/2 per symbol time, which is denoted by the modulation index h = 0.5 with h being defined as the number of π radians which the phase is changed per symbol time T.
With SFSK not only the phase but also the first derivative of the phase, so the frequency, is continuous. This improves side-band suppression.
If the phase shift with CPFSK is made such that smooth phase transitions between symbol intervals are realized, the phase change will no longer be constant over a symbol interval and one will commonly refer thereto as Continuous Phase Modulation (CPM). In view of the ever occurring phase changes over each symbol time, a special carrier reproducing circuit, for example as shown in Figure 1, has to be used for extracting the carrier signal from a received CPM-signal.
Examples of CPMs are Raised Cosine Modulation (RCM), having a self-explanatory name and the category of Correlative Phase Modulation (CORPM).
To achieve a further smoothing of the phase shift when using CORPM, the phase change in a symbol time also depends on the information of one or a plurality of preceding symbol times. Known CORPMs are the Tamed Frequency Modulation (TFM), Correlative Phase Shift Keying (CORPSK) and Gaussian Modified Shift Keying (GMSK).
Of all CPM-types only those types are of importance to the invention wherein at fixed points of time substantially fixed phase values occur which can be denoted as a whole fraction of 2π, for example 2π/2 with n = ±1, ±2, ...
These types of modulations will further be indicated as CPM with a rational modulation index h. If a carrier signal modulated thus is applied to input terminal 1 of the receiver comprising the carrier-reproducing circuit 10 represented in Figure 1, the signal is applied to a multiplier 3 in the conventional manner after being filtered in a band-pass filter 2 which will further be referred to as a third bandpass filter. Such a multiplier 3 comprises a non-linear element by means of which the modulated carrier signal, having a factor M and preferably being equal to 2/h, is multiplied for achieving a desired harmonic. As a result of the multiplication the phase variations of the modulated carrier signal end up in the same phase (modulo 2π). In known carrier-reproducing circuits for CPM modulated signals the thus obtained signal is directly applied to a first bandpass filter 4 having a small bandwidth, for example in the order of 1/100^th f_IN, wherein f_IN represents the frequency of the received signal, for eliminating all undesired harmonics, after which the desired harmonic of the carrier signal is divided in a divider 5 having dividend N. The thus reproduced carrier signal is applied to a coherent detector 6 likewise connected to the third wide-band filter 2 for coherent demodulation of the filtered input signal.
Figures 2a to 4b show several examples of CPM-signals having a substantially rational modulation index. Figure 2a shows therein the impulse response g(t) of a pulse applied to a modulator (not shown) in a transmitter for RC-signals. The time is plotted along the horizontal axis in units of the symbol time T and the amplitude is plotted along the vertical axis in units of T⁻¹. This Figure shows that the impulse response for RC-modulated signals represented therein has a length of 2T and is therefore a CORPM-signal indicated by 2RC. Figure 2b depicts in a waveform the phase shift of a carrier signal continuously phase-modulated by means of this impulse response. Such a waveform represents any occurring phase variations as a function of time of a carrier signal modulated by an arbitrary data signal. Along the vertical axis is plotted the phase in units of whereas along the horizontal axis the time is plotted in units of symbol time T.
As is shown in these Figures the phase at the instants mT, with m = 0, 1, 2, ..., has a specific value equal to one of the values nHπ, with n = 0, ± 1, ± 2, ... etc. The fact that the phase substantially has fixed values at the afore mentioned instants is caused by the surface under the impulse response curve g(t) and hence the modulation index having the value equal to h due to the dimensioning of the premodulation filter, so that the phase per time T can exactly change by the values of hπ. In most types of CPM modulation the phases at the instants mT have the constant unique values of (0 + 2π n) only after multiplication of the modulated carrier signal by the factor 2/h in the multiplier 3. An example of an exception to this is formed by TFM, wherein a substantially constant unique value ((±2n+1)π) occurs not only with a factor eight but also with a factor four, as will be explained hereinafter.
Figure 3a shows the impulse responses g(t) and Figure 3b shows the relevant waveform for CORPSK (2-3, 1+D) in solid lines and for duobinary CPM in dashed lines. In the indication (2-3, 1+D) the numeral 2 points out that the information signal applied to the premodulation filter is binary, the numeral 3 that the output signal of the premodulation filter is trinary for modulating the carrier signal with three different phase values, and 1+D that the successive information bits are correlated because in the premodulation filter the instantaneous standardized information bit, indicated by 1, has to be added to the preceding information bit delayed over a period time T indicated by the D of Delay.
In Figure 3a the area underneath the impulse response curve g(t) has the value of 1/2 so that all branches of the waveform pass through the fixed phase points n π/2 at the instants nT. Multiplication in multiplier 3 (Figure 1) by 2/h = 4 brings the phase at the aforementioned instants back to 0 ± 2 πn in the desired harmonic.
As a last example Figure 4a shows the impulse response g(t) and Figure 4b shows the associated waveform for TFM in solid lines whilst for TFSK (Tamed Frequency Shift Keying) the dashed lines are used. Here too the phase of the carrier signal turns out to have exact values at the end of the symbol intervals, in this case equal to n π/4. Consequently, a multiplication by a factor 8 is required in the multiplier (Figure 1). However, Figure 4b shows that in the centre of the symbol intervals the phases always turn out to be close to the values of π/4 + nπ/2, where n is an integer. Multiplying this signal by a factor 4 in the multiplier 3 will now suffice for achieving the substantially fixed values π + 2πn at the aforementioned sampling points (1/2 + m)T in the desired harmonic signal.
Despite the fact that the phases at the aforementioned sampling points do not exactly have the values of π/4 + nπ/2, with respect to the fixed phase points at the sampling points mT, this is advantageous in that a reliable carrier reproduction is feasible by a mere multiplication by a factor 4.
For CPM-signals having a rational modulation index it may be concluded from the aforeshown examples that by a multiplication by an integer the phases of the carrier signal are incremented such that at certain sampling points a single virtually fixed phase value occurs in the desired harmonic signal, for example 0 ± 2 πn or π± 2π n, that is to say a carrier phase irrespective of the symbol pattern of a data series.
Extracting the carrier signal by supplying the output signal of the multiplier 3 directly to the narrow-band first bandpass filter 4 in the conventional way, have a sub-optimal result due to the not insignificant jitter in the reproduced carrier signal. This jitter effect occurs as many elements causing disturbance occur in the signal after the multiplier 3 in addition to the desired phase values at the aforesaid sampling points.
To reduce this jitter effect the carrier reproducing circuit 10 comprises a sampling means 7 inserted between the multiplier 3 and the first bandpass filter 4, which sampling means is controlled by a clock signal coming from a clock signal regenerator 8. This clock signal regenerator 8 is connected to the wide band filter 2 for regenerating an accurate clock signal in the known manner from the signal produced by the wide band filter 2. Such a clock signal regenerator is, for example, described in the aforementioned article entitled "Synchronization properties of continuous phase modulation" by T. Aulin and C.E. Sundberg, published in the Conference Papers of Globecom 82, Global Telecommunications Conference, Miami, 29 November - 2 December 1982.
The clock signal regenerator 8 further comprises a delay means, which is not shown, (for example a delay line) having such a delay that the delayed clock pulses occur at the sampling point desired for a specific CPM signal in each symbol interval. If the receiver has to be suitable for receiving several types of CPM signals, a single adjustable delay means can be used to advantage rather than several switchable delay means.
As the phase of the modulated carrier signal has the desired value only at one or in some cases at two points per symbol interval, it is essential that the phase be sampled only at these points. This implies that sampling by means of a sampling pulse wider than an instantaneous sampling pulse will have a detrimental effect on the accuracy of the reproduced carrier signal.
A further improvement is obtained by inserting a relatively wide second bandpass filter 9, preferably having a bandwidth in the order of f_IN, between the multiplier 3 and the sampling means 7, to isolate the spectrum of the desired harmonic.
The improvement of a single C.P.M. signal having a rational modulation index, more specifically a TFM-signal, achieved by sampling will now further be explained with reference to a table showing computer simulations.
In the first column of the table is shown the S/N-ratio of the TFM-signal applied to the reproducing circuit of Figure 1.
The three remaining columns of the table show the square root of the mean square value of the phase error Δφ of the reproduced carrier signal over the last 136 data symbols of 156 received data symbols as well as the acquisition time T_acq of the receiver, expressed in symbol times T, assuming acquisition if the phase error is smaller than 5°. More specifically, the first column of these three shows the aforementioned values for a multiplication factor N=4, without sampling the second column shows the values for a multiplication factor N=4 with instantaneous sampling in the centre of the symbol intervals and the third column shows the values for a multiplication factor N=8 with instantaneous sampling at the end of the symbol intervals.

From this table it can be concluded that with instantaneous sampling in the centre of the bit interval (N=4) a considerable improvement in suppressing phase jitter is achieved for TFM-signals compared with the cases where no sampling is used. More specifically so with S/N ratios which are most important in practice. At a S/N-ratio of 10 dB, for example, the improved phase jitter suppression is even 6 dB. With instantaneous sampling at the end (N=8) of the symbol intervals, where the phase is exactly equal to a multiple of π/4, the jitter suppression is far better than with instantaneous sampling in the centre (N=4) of the symbol intervals, more specifically, up to a S/N ratio better than approximately 12 dB. However, at a S/N ratio of 10 dB the high multiplication factor starts taking its toll and the phase jitter in the reproduced carrier signal will rise rapidly.
Indeed, for practical values of the S/N ratio the multiplication factor 4 with centre-sampling is to be preferred. The above-mentioned example will have shown that improvements similar to those shown for TFM signals will likewise occur for other CPM-signals having a rational modulation index.

Claims

A carrier reproducing circuit for extracting the carrier signal from a pulse-coded modulated carrier signal, this circuit comprising a clock pulse regenerator (8), a multiplier (3) having a multiplication factor N(N = 2, 3, ...) for generating an N^th harmonic from the modulated carrier signal, a first bandpass filter (4) coupled to the multiplier for isolating the aforesaid N^th harmonic, a sampling circuit (7) which is inserted between the multiplier (3) and the first bandpass filter (4) and which is connected to the clock pulse regenerator (8) for sampling the N^th harmonic under the control of a clock pulse signal originating form the clock pulse regenerator at no less than one predetermined sampling point per symbol time T of the data signal and a divider (5) having the dividend N which is connected to the first bandpass filter characterized in that the pulse coded modulated carrier signal is a pulse coded continuous phase modulated carrier signal having a rational modulation index and that the sampling circuit (7) provides in an instantaneous sampling of the N^th harmonic under the control of the clock pulse signal.
A carrier producing circuit as claimed in Claim 1, characterized in that a second bandpass filter (9) is inserted between the multiplier (3) and the sampling circuit (7), this band filter (9) having a bandwidth exceeding by far the bandwidth of the first bandpass filter (4).
A carrier reproducing circuit as claimed in Claim 1, characterized in that the clock pulse regenerator (8) comprises an adjustable delay means for being capable of reproducing a carrier wave for any type of continuous phase-modulated carrier signal having a rational modulation index.
A receiver comprising a carrier reproducing circuit for extracting the carrier signal from a pulse-coded modulated carrier signal, this circuit comprising a clock pulse regenerator (8), a multiplier (3) having a multiplication factor N(N = 2, 3, ...) for generating an N^th harmonic from the modulated carrier signal, a first bandpass filter (9) coupled to the multiplier for isolating the aforesaid N^th harmonic, a sampling circuit (7) which is inserted between the multiplier (3) and the first bandpass filter (9) and which is connected to the clock pulse regenerator (8) for sampling the N^th harmonic under the control of a clock pulse signal originating form the clock pulse regenerator at no less than one predetermined sampling point per symbol time T of the data signal and a divider (5) having the dividend N which is connected to the first bandpass filter characterized in that the pulse coded modulated carrier signal is a pulse coded continuous phase modulated carrier signal having a rational modulation index and that the sampling circuit (7) provides in an instantaneous sampling of the N^th harmonic under the control of the clock pulse signal.